Project Summary Although a quickly expanding catalogue of protein mutations have been discovered, including those that cause human diseases or confer drug resistance, mechanistic understanding and prediction of functional consequences of these mutations remain rather limited. There is a critical need to develop novel, multiscale computational frameworks that are rigorous and generalizable to help close the ever-increasing gap between phenotypic data and mechanistic knowledge about the mutations and to develop effective therapeutic strategies. My long-term research goals are two folds: (1) to unravel how a change in protein sequence ripples through various aspects of molecular and system-levels to a change in cellular function and cause human diseases or confer drug resistance; and (2) to translate learned mechanistic knowledge to effective drug-design strategies for human diseases and drug resistance. Toward these goals I propose to advance and combine the strengths of computational molecular biology (structural modeling and design for proteins or protein interactions) and computational systems biology (pathway/network-level topology and dynamics) and follow a novel, formal paradigm to develop methods that systematically identify mechanisms underlying various stages of consequence propagation (from conformation, molecular and system-level variables, to cellular functions, where intermediates can be skipped if necessary). In the forward direction of propagating consequences, I will determine, explain and understand mutational consequences at various levels with machine learning, causal analysis, and combinatorial optimization, starting with known or modelled changes at various levels. In the inverse direction of designing consequences, I will follow known or learned mechanistic hypotheses and develop efficient combinatorial optimization methods to design mutagenesis experiments or screen / design ligands for desired perturbations at various levels, which would test and refine corresponding hypotheses directly and translate mechanistic hypotheses to therapeutic strategies. Mechanisms underlying distal (potentially allosteric) mutations, multiple mutations within a protein or across proteins as well as systems pharmacology strategies will also be rigorously determined in the process. The expected outcomes from the innovative project will include two main components. The first will be both biophysical principles of proteins and integrative understanding of biomedical systems, which will be organized in a database for public access. The second will be translating mechanistic insights into therapeutic intervention strategies to treat diseases. In particular, targeted experiments (mutagenesis or ligand-binding) will be designed following known or discovered mechanistic hypotheses. This project will involve close collaboration with structural biologists and clinician-scientists who are active collaborators of the PI.